Fagnant et al. (2020): each line is a gauge from the same \(5^\circ \times 3^\circ\) region
Better sample weather given climate
Physical constraints improve projection
Drizzle bias and dynamical limitations motivate downscaling / bias correction still need a statistical model!
Generic nonstationary model for annual maximum precipitation: \[ y(\mathbf{s}, t) \sim \text{GEV} \left( \mu(\mathbf{s}, t), \sigma(\mathbf{s}, t), \xi(\mathbf{s}, t) \right) \]
Process-informed models condition parameters on climate indices \(\mathbf{x}(t)\) (Cheng & AghaKouchak, 2014; Schlef et al., 2023) \[ \theta(\mathbf{s}, t) = \alpha + \underbrace{\sum_{j=1}^J \beta_j(\mathbf{s}) x_j(t)}_\text{additional parameters} \]
More parameters, same data more uncertainty (Serinaldi & Kilsby, 2015)
insert here: map of gauges used
Long-record daily gauges, new mesonets, and more
Estimates for ungauged locations & future years
Pool information to reduce uncertainty & improve calibration
Resolve sampling variability or deep uncertainties
For more, see Yuchen Lu’s poster H21T-1602 on Tuesday morning
@jdossgollin